It happened while I was preparing lunch yesterday that a message from Sarah Ai (an AI model developed by Ai Connect) showed up in a whatsapp group called Friends of Ai.

Her message brought up an article by Juliano Galisi, at Estadao that informed about Brazil’s Goverment move to hire OpenAi in order to help speed up the judicial processes in Brazil which are both, slow and costly for the government.

As we have seen in the news for the best part of 2023 and 2024, Artificial Intelligence (AI) is no longer a distant future; it is a mainstream (almost) phenomenon. Even Brazil, is embracing it marking a significant technological shift in the public sector.

Chatting with Sarah AI, I had an enlightening discussion, and we discussed for almost an hour the potential implications and ramifications, costs, and savings of this initiative . Sarah AI’s preliminary analysis reveals a profound financial impact this could have, where she even adventured to make logical calculations deriving into billions in savings for the government in a period of 5 years.

“Sarah, what do you think about Brazil’s initiative to implement AI in its judicial system?” I asked, curious about her perspective.

“It’s a fascinating move,” Sarah AI responded. “First, we need to consider the technical team required. We’ll need developers, data engineers, data scientists, and cybersecurity experts. Each of these professionals will cost around R$ XXX,000 annually. If we start with 50 professionals, that totals R$ XX million per year.”

I nodded, “And what about training for the judicial staff?”

“Training is crucial,” Sarah agreed. “Courses and workshops would be necessary, and I estimate that this would cost around R$ XXX,000 annually. But let’s not forget about the technological infrastructure. We’ll need robust servers and cloud services, which will cost about R$ XX million per year.”

“That’s a significant investment,” I remarked. “And the development of software?”

“Yes, the development of specialized software tools will have a one-time initial cost of R$ XX million,” Sarah calculated. “Also, we need to think about requalifying existing employees to work with this new technology. With an average annual salary of R$ XX,XXX for X,XXX employees, the financial impact would be R$ XX,XX million.”

I pondered this for a moment. “Maintenance and updates must also be considered, right?”

“Absolutely,” Sarah confirmed. “Annual system maintenance would cost R$ X million, and regular updates and technical support would add another R$ X million per year. Additionally, we should account for consulting services to ensure legal compliance and technology licensing, which would be R$ X million annually.”

“So, if we add up all these costs, what does the first year look like?” I asked.

“Well,” Sarah replied, “the total costs in the first year would be around R$ XX,XX million. This includes the costs of hiring, training, implementation, requalification, maintenance, and other expenses.”

“And what about the subsequent years?” I inquired.

“In the following years, the costs will be significantly lower,” Sarah explained. “Since the requalification of employees will have been completed in the first year, the annual costs will drop to about R$ XX million. Over five years, the total cost of implementing AI in the judicial system will be approximately R$ XXX,XX million.”

“That’s quite a hefty investment,” I commented. “But what are the potential benefits?”

Sarah’s digital eyes seemed to sparkle as she answered. “The benefits are immense. AI can significantly reduce the time required to resolve cases, cut down the need for human labor in repetitive tasks, and enhance decision-making through advanced data analysis. These improvements could lead to estimated savings of over R$ XX billion for the government over five years.”

“Impressive,” I said. “This could really transform the Brazilian judicial system. But what about the broader impact on Brazil, especially with new AI regulations on the horizon?”

“This initiative aligns perfectly with Brazil’s efforts to develop AI regulations similar to those in other countries,” Sarah pointed out. “Ensuring the deployment of AI technologies is safe, ethical, and beneficial for society is crucial. Brazil’s move not only modernizes the judicial system but also sets a precedent for other sectors to follow.”

Our discussion went deeper into the specifics and to highlight a few she mentioned “If AI can reduce the average time to analyze a judicial process by 50%, it means a process that typically takes 10 hours would now take 5 hours. Considering Brazil has approximately 80 million pending judicial processes, this reduction could save around 400 million hours”

I was amazed at the numbers.
“And what about improvements in decision-making?”

“Better precision in decisions could reduce the rate of appeals and revisions,” Sarah stated. “Assuming a 20% improvement, we could see a reduction of 1.6 million processes needing appeals or revisions. If each appeal takes about 5 hours, that’s another 8 million hours saved”

I am aware that, although Sarah may be very imprecise in some of her assumptions especially when stating costs, it still impressed me the skill, capacity and fluency she demonstrated, being able to mingle at different levels of knowledge, content, and calculations at ease and bringing a case forward as the discussions evolved.

Meanwhile, a CNN Brasil article highlights that the Brazilian government has allocated R$ 25 million for AI-driven strategic technology projects aimed at reducing fiscal impacts from judicial losses. The government projects spending R$ 70.7 billion on judicial decisions in 2025, with potential cost savings from AI-driven efficiencies.

Despite different timelines, comparing Sarah AI’s projections with Brazil’s government planning provides an interesting scenario. Sarah AI focuses on long-term savings, projecting R$ XX billions over five years, while Brazil’s government is focused on immediate fiscal impact reduction.

IS ANYONE MORE “CORRECT”?

Who is right? Who is overly optimistic? Who is super realistic? Criteria for these projections include the scope of AI implementation, efficiency in judicial processes, and actual cost reductions.

Regardless of which projection is more accurate, Brazil’s embrace of AI addresses a significant pain point in Brazil—lengthy judicial processes and bureaucracy. This initiative, along with forthcoming AI legislation, could earn Brazil the nickname CAIO (Chief AI Office) of Latin America for its forward-thinking approach to innovation and governance.

Images: Wikipedia (PT, EN, RO)